Super-resolution imaging of synaptic proteins for the precise localization within the

4.5 Methods

5.3.2 Super-resolution imaging of synaptic proteins for the precise localization within the

2010; Nouvian et al, 2011), the expression of syntaxin 12/13 and VAMP4 have not been investigated much.

SNAP29, another SNARE protein, which is similar to the synaptic SNARE SNAP25, is also located to the golgi network and interacts with syntaxin 6 (Wong et al, 1999), but has also been reported to localize ubiquitously to intracellular membranes (Steegmaier et al, 1998).

Additionally, SNAP29 has been associated with autophagy (Itakura et al, 2012; Morelli et al, 2014). Whether SNAP29 is expressed in IHCs is not known.

5.3.2 Super-resolution imaging of synaptic proteins for the precise localization within the ribbon synapse

After having selected target proteins known to be involved in the synaptic vesicle recycling process and the appropriate antibodies to detect them, I proceeded with the staining and imaging of those target proteins using super-resolution microscopy. For STED imaging, I stained cryosections of the Organ of Corti for the POI and additionally for the IHC marker vGlut3 and the ribbon synapse marker CtBP2 or ribeye, which both label the ribbon. vGlut3

124 staining was only used to identify IHCs and thus imaging in confocal resolution was sufficient. The stainings of the synaptic ribbon and the protein of interest were imaged in STED mode, using the laser line providing better resolution (640nm) for the POI (Figure 2A).

In order to provide an average distribution of the protein of interest within the ribbon synapse of the hair cell, I stained and imaged at least 4 cryosections per POI independently, each containing multiple ribbon synapses. The images of each synapse were aligned and rotated according to a model (Figure 2B), so the orientation of each synapse was the same and the staining signal could be averaged over every synapse of one cryosection (Figure 2C) and finally over all synapses from all cryosections (Figure 2D). The resulting average distribution maps were generated for all target proteins (Figure 3).

Figure 2 STED imaging for determining the spatial organization of synaptic proteins in the IHC ribbon synapse.

A) Example images of IHC ribbon synapses stained for the protein of interest clathrin (red). Samples were always stained for a ribbon synapse marker (ribeye or CtBP2; blue) and an IHC marker (vGlut3;

green). The IHC marker was imaged in confocal mode, in order to locate the cells. Synapse marker as well as the POI were imaged in STED mode for precise localization of the ribbon and the protein of interest in relation to the ribbon. Scale bar = 2 m. B) For the analysis of the spatial orientation of each POI, images of each ribbon synapse area were oriented according to a superficial model (B). This way all synapses were centered with the IHC cytosol (labeled by the IHC marker) facing the upper half of the image (white half circle in the model). C) Staining signals of each marker could be averaged over all synapses from one sample. D) Averaging signals over all samples resulted in average distribution maps. The color map indicates the intensity of the signal (from blue – low to red – high). Scale bar in C and D 1 m.

From the average distribution maps, one can see that some proteins (amphiphysin, CSP and VAMP4) seem to have a very localized distribution with a few hotspots within the image area (Figure 3). Most proteins though seem to be distributed rather ubiquitously within the area centered around the ribbon. A few proteins show a distribution, which seems almost random within the whole area of analysis, not even matching the IHC cytosol labeling (SCAMP1, synaptogyrin, synaptotagmin 2, synaptotagmin 7, VAMP3, synaptojanin). Whether those signals originate from true protein distributions remains to be carefully discussed.

125 Figure 3 Average distribution maps of each POI in relation to the ribbon synapse marker and IHC marker.

Staining signals of the ribbon marker, the IHC marker and the protein of interest were averaged over all detected ribbon synapses, resulting from the imaging of 4 – 8 IHC samples (cryosections) per POI.

The color map indicates the intensity of the signal (blue – low; red – high) and is not scaled equally between different POI samples. Image dimensions are 3 m x 3 m. Some proteins of interest seem

126 to be very localized (e.g. Amphiphysin, CSP or VAMP4), whereas most of the investigated proteins seem to be rather dispersed within the boundaries of the IHC marker.

During the imaging process I had the impression that for some proteins (CSP, endophilin and vATPase), a substantial part of the signal was localized outside the IHC, but close to the synaptic ribbons. These signals might originate from the proteins expressed in efferent synapses. In order to confirm this impression, I repeated the stainings of those POIs with a co-staining of synaptophysin, which is known to be expressed in synapses of efferent neurons (Gil-Loyzaga & Pujol, 1988; Figure 4). Strong co-localization of synaptophysin with major parts of the signal from CSP, endophilin and vATPase stainings were detected. This confirms that the strong labeling signal outside of the IHCs originate from efferent synapses.

Nevertheless, less bright signal of those three POIs was also detected within the area of IHC labeling, which indicates that these proteins are also expressed in IHCs, although to a lower amount. This assumption is supported by previous studies, detecting these proteins in IHCs (Uthaiah & Hudspeth, 2010; Eybalin et al, 2002; Hickox et al, 2017; Scheffer et al, 2015; Liu et al, 2014b; Kroll et al, 2019), as discussed in section 6.3.

127 Figure 4 Co-immunostaining of synaptophysin and the POIs CSP, endophilin and vATPase.

A) Example images of the stainings used for protein localization. Bright labeling of the proteins of interest CSP, endophilin and vATPase (red) were found outside of the IHC marker area (green), but in the vicinity of ribbon synapses (arrowhead). B) Co-staining of the POIs with synaptophysin as marker for efferent synapses. Strong co-localization of each target protein with synaptophysin can be observed (open arrowheads). Less bright signal of CSP, endophilin and vATPase stainings could also be found within the IHCs (arrows). Scale bar = 1 m.

128 5.3.3 Protein copy number estimation using a comparative imaging approach (CosiQuant) In order to get an estimate for the copy number of each investigated protein in the IHC ribbon synapse, I used the comparative imaging technique CosiQuant that was introduced in chapter 04. To recapitulate, this method is based on the comparison of signals between a sample of interest and synaptosome preparations that have been immunostained in parallel for the same protein of interest. I have stained cryosections of the Organ of Corti and synaptosomes (the same preparations as in chapter 04) for all 19 POIs using the same protocol and imaged the samples in parallel, using a confocal microscope. Small adjustments had to be made in comparison to the protocol in chapter 04. The synaptosome samples were still immunostained for the two synapse markers synaptophysin and bassoon in addition to the POI, but IHCs were stained for the IHC marker vGlut3 and the ribbon marker CtBP2 or ribeye instead (Figure 5 and 6). The staining procedure for the POI was kept identical between IHC samples and synaptosomes. The signal of the two markers were used for the automatic identification of synapses and the intensity of the POI label was measured in that area. For each POI, the amount of non-specific background signal was assessed by control stainings, using only secondary antibodies, without the primary antibody. These control stainings were performed and imaged under the same conditions as the IHC samples and the synaptosome samples. Thus, mean intensity values for the POIs could be corrected for background signal. To account for differently sized areas of measurement, influencing the comparison of copy numbers between the different samples, the mean signal intensity was calculated for the number of pixels measured. By doing so, the intensities could be compared between different areas measured in IHCs and synaptosomes. The intensity for each immunostained POI in IHCs was then expressed as fold over synaptosome signal (Figure 6B) and from this ratio and the known copy numbers of the POIs in synaptosomes, I estimated the protein copy numbers in the IHC synapse (Figure 6B).

Although multiple independent stainings were imaged and analyzed to account for variability in the staining and imaging procedure, the mean intensities of the stained POIs still exhibit substantial variation (Figure 6B). This variability might represent differential staining efficiency or real variation in protein expression between different synapses (see section 5.4. for further discussion on this). As for the estimated protein copy numbers, one can sort the proteins into different abundancy categories for a better overview. By far the most abundant protein seems to be the small GTPase rab3, which resides on the synaptic vesicles. An intermediate abundancy can be seen for the endocytosis proteins clathrin, AP180, dynamin and amphiphysin. The proteins vATPase, SCAMP1 and endophilin show an intermediate to low copy number. The rest of the proteins seem to be present in low copy numbers, amongst these are all tested SNARE proteins, synaptojanin and the exocytosis proteins synaptotagmin 2, synaptotagmin 7, CSP and synaptogyrin.

129 Figure 5 Protein copy number estimation by comparison of immunostaining signals.

Example images of IHC cryosections (IHC) and synaptosomes (Syn) stained and imaged in parallel for the same POI and different sample markers (synaptophysin for synaptosomes and vGlut3 for IHCs) and synapse markers (bassoon for synaptosomes and ribeye or CtBP2 for IHCs). Copy numbers for the POIs can be estimated from comparing signal intensities between biochemically characterized synaptosomes and IHC samples based on the CosiQuant method established previously in chapter 04. Images of synaptosomes and IHCs are scaled equally for the same POI stainings, images are not corrected for background signal here. Scale bar = 1 µm.

130 Figure 6 Protein copy number estimation by comparison of immunostaining signals (continued).

A) Example images of IHC cryosections (IHC) and synaptosomes (Syn) stained and imaged in parallel for the same POI and different sample markers (synaptophysin for synaptosomes and vGlut3 for IHCs) and synapse markers (bassoon for synaptosomes and ribeye or CtBP2 for IHCs). Images of synaptosomes and IHCs are scaled equally for the same POI stainings, images are not corrected for background signal here. Scale bar = 1µm. B) Copy numbers for the POIs (lower panel) can be estimated from comparing mean signal intensities of the POI (upper panel) between biochemically characterized synaptosomes and IHC samples based on the CosiQuant method established previously in chapter 04. Mean intensities of the POI in IHCs are background substracted signals, calculated for the measured area and expressed as fold over synaptosome signal. Copy numbers for VAMP3 are currently under evaluation by dot plot analysis, since the copy numbers in Wilhelm et al, 2014 were obtained using a different antibody. Error bars represent the SEM. N = 3 – 7 independent stainings containing multiple images.

131 5.3.4 Incorporation of protein copy numbers and spatial organization into a 3-dimensional model

After having determined the localization and estimated the copy numbers of the selected synaptic proteins, I aimed to combine this data in a 3-dimensional model of the hair cell ribbon synapse. Therefore, I reconstructed a ribbon synapse from electron microscopy images (Figure 7) taken from serial sections of a wildtype mouse cochlea (P12). This electron microscopy data was kindly provided by Prof. Dr. Carolin Wichmann (University of Göttingen Medical Center, Göttingen, Germany). Serial sections of 60 nm thickness contained the complete structure of the ribbon and adjacent synaptic vesicles (Figure 7A). Manual tracing of structures and organelles like the plasma membrane, active zone, synaptic ribbon, synaptic vesicles and vacuoles resulted in a 3-dimensional model of the ribbon synapse (Figure 7B), which can be used as a basis for further modelling of the proteins within that synapse area.

Figure 7 3-dimensional reconstruction of an IHC ribbon synapse.

A) Electron microscopy images of serial sections through an IHC ribbon synapse served as a basis for the 3-dimensional reconstruction of such a ribbon synapse. A few example sections are shown here, containing the plasma membrane with the active zone (red arrowhead), the ribbon (white arrow), synaptic vesicles (open arrowhead) and vacuoles (black arrowheads). Scale bar = 100 nm. B) 3-dimensional model of the ribbon synapse, consisting of the plasma membrane (brown) with the active zone (red), the ribbon (purple) and synaptic vesicles (grey). Vacuoles were traced and reconstructed as well, but are not shown here for better visibility.

In order to model the previously investigated proteins in correct abundancies, the estimated protein copy numbers needed to be calculated according to the reconstructed space of the ribbon synapse (model area in Table 2). Protein copy numbers were calculated from the previously estimated numbers within the specific imaging areas (measured area in Table 2), so they match the reconstructed area taken from the EM data.

132 Table 2 Mean intensities and estimated protein copy numbers for synaptic proteins in IHCs

Mean intensities are expressed as fold over synaptosome intensities; SEM = standard error of the mean. Measured area refers to the image area, on which the copy number estimation was based.

Model area refers to the 3-dimensional model area based on electron microscopy data.

POI mean intensity SEM protein copy number (measured area)

protein copy number (model area)

Amphiphysin 23.0777 11.1836 27559.3893 10218.7246

AP180 20.1603 8.7207 75327.1258 27930.4866

Clathrin 26.5421 7.4984 106168.457 39366.12

CSP 4.0508 1.5701 3812.5633 1413.6574

Dynamin 12.7938 6.4514 29763.5739 11036.0124

Endophilin 6.0927 2.0502 15380.5129 5702.9284

Rab3 15.028 5.6671 283226.404 105017.299

SCAMP1 13.2337 4.2182 19314.5435 7161.6246

SNAP29 5.2442 2.7598 406.2708 150.6408

Synaptogyrin 3.4685 0.599 6433.448 2385.4532

Synaptojanin 9.4347 2.553 3443.6777 1276.8786

Synaptotagmin2 14.7053 9.6276 4371.5767 1620.9336

Synaptotagmin7 9.1195 3.8221 1665.5763 617.5778

Syntaxin1213 3.5105 16,594 554.0609 205.4398

Syntaxin16 41.877 14.3859 3822.1174 1417.2

Syntaxin6 67.0355 34.6118 8156.2093 3024.2346

VAMP3 4.518 1.1203 * *

VAMP4 8.267 2.7127 831.5775 308.34

vATPase 28.5859 11.4616 21221.3146 7868.6348

*copy numbers are currently being evaluated by dot plot analysis, since the copy numbers in Wilhelm et al, 2014 were obtained using a different antibody

To get a better graphical overview of the data obtained from this study, some preliminary modelling of the proteins within the reconstructed synapse was done. For this, the proteins were placed in a 3-dimensional coordinate system according to their estimated abundancy and localization determined by STED microscopy (Figure 8 and 9). The proteins are represented as simple green crosses until more advanced modelling can be done in cooperation with Burkhard Rammner (University of Göttingen Medical Center, Göttingen, Germany), who already modelled the presynaptic bouton in Wilhelm et al, 2014. For now, knowledge of proteins being associated to certain organelles (like the synaptic vesicle) has been neglected for simplicity purposes. Since the proteins I investigated in this project were based on the available protein numbers from the Wilhelm et al, 2014 study, the basic models of those proteins already exist and only need to be adjusted in number and location according to my data and the literature for a more advanced model of the ribbon synapse.

133 Figure 8 Preliminary modelling of the 3-dimensional synaptic space of the IHC ribbon synapse, containing the investigated proteins.

Tracings of the plasma membrane (brown), active zone (red), synaptic ribbon (purple) and synaptic vesicles (grey) per EM section are represented in a 3-dimensinal coordinate system. Each investigated POI (green) was placed in that synaptic space according to the copy number estimation and super resolution localization. For display purpose, the first 2 panels show an ‘empty synapse’, containing only the membrane, active zone, ribbon and vesicles. Thereafter, each POI is shown separately in that model of the ribbon synapse.

134 Figure 9 Preliminary modelling of the 3-dimensional synaptic space of the IHC ribbon synapse, containing the investigated proteins (continued).

135 containing only the membrane, active zone, ribbon and vesicles. Thereafter, each POI is shown separately in that model of the ribbon synapse.

5.4 Discussion

Due to its ability of very precise temporal stimulus-transmission coupling and at the same time maintenance of sustained transmission, the IHC ribbon synapse is an interesting subject for investigating the mechanism of synaptic vesicle recycling. Vesicle recycling needs to be very efficient in these synapses in order to ensure synaptic transmission over a long time of stimulation without substantial fatigue. This is why it has been tried to unravel the details of this process in the past using various techniques, from electrophysiology to mass spectrometry. Nevertheless, until now the exact protein composition of the synaptic vesicle recycling machinery is unclear. Since this protein composition seems to differ from conventional synapses (Gil-Loyzaga & Pujol, 1988; Safieddine & Wenthold, 1997; Mandell et al, 1990; Roux et al, 2006; Michalski et al, 2017; Wenthold et al, 2002), it is important to have a clear picture of the molecular organization of the ribbon synapse in IHCs, before investigating detailed functions and interactions of proteins by genetic alterations and biochemical assays. In this study, I intend to provide such an overview of the IHC ribbon synapse, containing detailed information about the spatial orientation and abundancy of synaptic proteins within the presynaptic area. I achieved to do so by investigating the presence of 19 synaptic proteins and their exact location in relation to the ribbon by super-resolution STED imaging. Fluorescence microscopy-based localization of ribbon synapse proteins has neither been done in such detail nor for more than 1 – 2 proteins per study before. Implementing an improved fixation method and combining it with a precise co-localization of synaptic proteins and the ribbon by 2-colour STED microscopy, enabled me to conduct such a detailed investigation. Furthermore, the application of CosiQuant, a method to estimate protein copy numbers in samples which are difficult to investigate by biochemical techniques, allowed me to assess the abundancy of 19 proteins in the IHC ribbon synapse.

It has to be noted that CosiQuant is a semi-quantitative method and only provides estimates for protein copy numbers. This is especially relevant for the 19 synaptic proteins investigated in this study, since immunostaining signals of these proteins showed substantial variability (see Figure 6 in subsection 5.3.3). It is difficult to determine the cause of this variation in staining signals. It might be due to varied labeling efficiencies of the target protein by the primary and secondary antibody. This could be caused by different accessibilities of the epitope and different penetration efficiencies of the antibodies, depending on how the Organ of Corti was dissected and sectioned. Although at least penetration should not be a major problem in a 10 µm cryosection of the organ, since whole Organs of Corti can be stained using similar staining protocols. Increased variability in fixation quality between the samples or target proteins using the new fixation method (glyoxal) can also be neglected, since variation of the immunostaining signal seems to be independent of which fixative was used. Alteration in imaging parameters (due to e.g. fluctuations in laser intensities of the

136 setup) could potentially also contribute to the variation in measured signal intensities. This can however be neglected here, since synaptosome samples and IHC samples were always imaged in parallel on the same day. The high variance of the immunostaining intensity of the investigated proteins might also represent real variability between ribbon synapses. Despite dissecting the Organ of Corti in a buffer without Ca²⁺, there were no additional measures to inhibit or stimulate the synapses. Thus, variation in protein labeling signal might be due to different amounts of protein at the ribbon depending on the activity status, similar to the principle of different amounts of postsynaptic receptors in synaptic plasticity mechanisms at neuronal synapses (Fernandes & Carvalho, 2016; Pozo & Goda, 2010). Furthermore, ribbon synapses have been described to show high heterogeneity in morphology depending on their location within the IHC (modiolar side vs. pillar side) and along the tonotopic axis (Michanski et al, 2019; Safieddine et al, 2012). For imaging experiments, I only took the apical turn of the Organ of Corti to minimize the variability between synapses from different tonotopic positions, but I was not able to distinguish between ribbon synapses from modiolar and pillar sides. Thus, the variation in signal intensity for a certain protein might reflect the variation of protein expression at different ribbon synapses, according to the morphological parameters of the synapse.

Regardless of what causes the variation in immunostaining intensity, the estimates for protein copy numbers based on this signal is still valuable information, providing an overview of protein abundance at the IHC ribbon synapse. Amounts of the investigated proteins relative to each other and in comparison to protein amounts in conventional synapses now provide a starting point for evaluating the efficiency of the synaptic vesicle recycling process. It is possible to determine rate-limiting factors for this process, depending

Im Dokument The molecular anatomy of synaptic vesicle recycling at the hair cell ribbon synapse (Seite 131-0)